Research Seminar of the Statistical Science Research Core

1
2

3

Day & Time
 
29th October 2014, 15:00~
Venue
 
Room C709, Building of the Faculty of Integrated Arts and Sciences
Lecturer
 
Dr. Hideitsu HINO (Assistant Professor, Faculty of Engineering, Information and Systems, University of Tsukuba)

Presentation Title
Nonparametric Estimation of Differential Entropy and Its Application
Outline
The Shannon’s differential entropy has a broad range of applications, such as independent comportment analysis, dimension reduction, and change detection. In exploratory data analyses, it is not desirable to assume some distribution of the given data, and nonparametric estimation as entropy estimate can be effective for these situations.
In this lecture, I will introduce nonparametric estimate from the weighted data, and its application to information theoretic clustering. Then, I will discuss a new entropy estimate based on the quadratic expansion of the probability mass function and regression analysis.